KLOUDY / models
Models
You do not sign up at five model vendors to try a routing stack. Kloudy routes cheap models first and only climbs the Nemotron ladder when the job is hard. Your IDE model stays yours.
Product overview. Public-page reading and KLDY preparation are available here. The connector, account actions and native apps described below are not connected to this site.
Less setup. More useful work.
For IDEs and bots: one CLI, one MCP, tools appear when needed.
Find with SEO/GEO, read FHIR for medical data or KLDY for other information, then select one scoped Toolbox tool when action is needed.
Passive local reads use no model calls. Narrowing 100 candidate definitions to 10 would mean 90% fewer definition fetches in that example; this is an illustration, not measured token savings.
Nemotron cost basis
The shared model direction uses the Nemotron ladder, from small models for simple tasks to larger models for complex work. Exact model availability and prices must come from the verified provider catalog and a dated rate card. AWS hosts this browser and public reader; AWS hosting does not mean a Bedrock model is used. A public URL read does not make a model call.